High Performance Reconfigurable Hardware Acceleration on Neutron Transport Computation Based on AGENT Methodology

2010 ◽  
Author(s):  
Shanjie Xiao ◽  
Tatjana Jevremovic
Keyword(s):  
High Performance ◽  
Neutron Transport ◽  
Three Dimensional ◽  
Hardware Acceleration ◽  
Reconfigurable Hardware ◽  
Reactor Physics ◽  
Von Neumann ◽  
Acceleration Techniques ◽  
High Efficient ◽  
Field Programmable

A high performance hardware acceleration coprocessor built on field programmable arrays (FPGAs) is designed to accelerate neutron transport computation for three dimensional whole reactor cores. The acceleration coprocessor is designed based on the reconfigurable computation techniques and adopts the dataflow-driven non von Neumann architecture for high efficient parallel computation. The hardware acceleration coprocessor supports much more intensive available computation power compare with the same-era CPUs, and is compatible with existing software acceleration methods. It reaches about 20 times speed up in simulation validations. It is the first time that the reconfigurable hardware acceleration techniques are used to improve the computational efficiency of the reactor physics and neutron transport simulations.

scholarly journals High Level Design of a Flexible PCA Hardware Accelerator Using a New Block-Streaming Method

Electronics ◽  
2020 ◽  
Vol 9 (3) ◽  
pp. 449
Author(s):  
Mohammad Amir Mansoori ◽  
Mario R. Casu
Keyword(s):  
High Performance ◽  
Principal Component ◽  
Hardware Acceleration ◽  
Design Flow ◽  
Hardware Accelerator ◽  
Field Programmable ◽  
Point Solution ◽  
Active Research ◽  
High Level ◽  
Many Core

Principal Component Analysis (PCA) is a technique for dimensionality reduction that is useful in removing redundant information in data for various applications such as Microwave Imaging (MI) and Hyperspectral Imaging (HI). The computational complexity of PCA has made the hardware acceleration of PCA an active research topic in recent years. Although the hardware design flow can be optimized using High Level Synthesis (HLS) tools, efficient high-performance solutions for complex embedded systems still require careful design. In this paper we propose a flexible PCA hardware accelerator in Field-Programmable Gate Arrays (FPGA) that we designed entirely in HLS. In order to make the internal PCA computations more efficient, a new block-streaming method is also introduced. Several HLS optimization strategies are adopted to create an efficient hardware. The flexibility of our design allows us to use it for different FPGA targets, with flexible input data dimensions, and it also lets us easily switch from a more accurate floating-point implementation to a higher speed fixed-point solution. The results show the efficiency of our design compared to state-of-the-art implementations on GPUs, many-core CPUs, and other FPGA approaches in terms of resource usage, execution time and power consumption.

Performance, Accuracy and Efficiency Evaluation of a Three-Dimensional Whole-Core Neutron Transport Code AGENT

2006 ◽  
Author(s):  
Tatjana Jevremovic ◽  
Mathieu Hursin ◽  
Nader Satvat ◽  
John Hopkins ◽  
Shanjie Xiao ◽  
...  
Keyword(s):  
Monte Carlo ◽  
Neutron Transport ◽  
Three Dimensional ◽  
Open Architecture ◽  
Coarse Mesh ◽  
Research Reactors ◽  
Acceleration Techniques ◽  
Performance Accuracy ◽  
Transport Code ◽  
Neutron Propagation

The AGENT (Arbitrary GEometry Neutron Transport) an open-architecture reactor modeling tool is deterministic neutron transport code for two or three-dimensional heterogeneous neutronic design and analysis of the whole reactor cores regardless of geometry types and material configurations. The AGENT neutron transport methodology is applicable to all generations of nuclear power and research reactors. It combines three theories: (1) the theory of R-functions used to generate real three-dimensional whole-cores of square, hexagonal or triangular cross sections, (2) the planar method of characteristics used to solve isotropic neutron transport in non-homogenized 2D) reactor slices, and (3) the one-dimensional diffusion theory used to couple the planar and axial neutron tracks through the transverse leakage and angular mesh-wise flux values. The R-function-geometrical module allows a sequential building of the layers of geometry and automatic submeshing based on the network of domain functions. The simplicity of geometry description and selection of parameters for accurate treatment of neutron propagation is achieved through the Boolean algebraic hierarchically organized simple primitives into complex domains (both being represented with corresponding domain functions). The accuracy is comparable to Monte Carlo codes and is obtained by following neutron propagation through real geometrical domains that does not require homogenization or simplifications. The efficiency is maintained through a set of acceleration techniques introduced at all important calculation levels. The flux solution incorporates power iteration with two different acceleration techniques: Coarse Mesh Rebalancing (CMR) and Coarse Mesh Finite Difference (CMFD). The stand-alone originally developed graphical user interface of the AGENT code design environment allows the user to view and verify input data by displaying the geometry and material distribution. The user can also view the output data such as three-dimensional maps of the energy-dependent mesh-wise scalar flux, reaction rate and power peaking factor. The AGENT code is in a process of an extensive and rigorous testing for various reactor types through the evaluation of its performance (ability to model any reactor geometry type), accuracy (in comparison with Monte Carlo results and other deterministic solutions or experimental data) and efficiency (computational speed that is directly determined by the mathematical and numerical solution to the iterative approach of the flux convergence). This paper outlines main aspects of the theories unified into the AGENT code formalism and demonstrates the code performance, accuracy and efficiency using few representative examples. The AGENT code is a main part of the so called virtual reactor system developed for numerical simulations of research reactors. Few illustrative examples of the web interface are briefly outlined.

Electronic Hardware for Fuzzy Computation

2010 ◽  
pp. 1-30 ◽  
Author(s):  
Koldo Basterretxea ◽  
Inés del Campo
Keyword(s):  
Data Mining ◽  
Reconfigurable Computing ◽  
High Performance ◽  
Reconfigurable Hardware ◽  
Dynamic Partial Reconfiguration ◽  
Field Programmable ◽  
Analog Approach ◽  
Programmable Gate Arrays ◽  
System On Programmable Chip ◽  
Electronic Hardware

This chapter describes two decades of evolution of electronic hardware for fuzzy computing, and discusses the new trends and challenges that are currently being faced in this field. Firstly the authors analyze the main design approaches performed since first fuzzy chip designs were published and until the consolidation of reconfigurable hardware: the digital approach and the analog approach. Secondly, the evolution of fuzzy hardware based on reconfigurable devices, from traditional field programmable gate arrays to complex system-on-programmable chip solutions, is described and its relationship with the scalability issue is explained. The reconfigurable approach is completed by analyzing a cutting edge design methodology known as dynamic partial reconfiguration and by reviewing some evolvable fuzzy hardware designs. Lastly, regarding fuzzy data-mining processing, the main proposals to speed up data-mining workloads are presented: multiprocessor architectures, reconfigurable hardware, and high performance reconfigurable computing.

scholarly journals Martian dust devils detector over FPGA

2011 ◽  
Vol 1 (1) ◽  
pp. 1-29
Author(s):  
E. de Lucas ◽  
M. J. Miguel ◽  
D. Mozos ◽  
L. Vázquez
Keyword(s):  
High Performance ◽  
Hardware Implementation ◽  
Small Volume ◽  
Space Missions ◽  
Reconfigurable Hardware ◽  
Dust Devil ◽  
Dust Devils ◽  
Gate Arrays ◽  
Field Programmable ◽  
Programmable Gate Arrays

Abstract. Digital applications that must be on-board of space missions must accomplish a very restrictive set of requirements. These include energy efficiency, small volume and weight, robustness and high performance. Moreover these circuits can not be repaired in case of error, so they must be reliable or provide some way to recover from errors. These features make reconfigurable hardware (FPGAs, Field Programmable Gate Arrays) a very suitable technology to be used in space missions. This paper presents a Martian dust devil detector implemented on a FPGA. The results show that a hardware implementation of the algorithm present very good numbers in terms of performance compared with the software version. Moreover, as the amount of time needed to perform all the computations on the reconfigurable hardware is small, this hardware can be used more of the time to realize other applications.

scholarly journals MPACT VERIFICATION WITH MAGNOX REACTOR NEUTRONICS PROGRESSION PROBLEMS

2021 ◽  
Vol 247 ◽  
pp. 10031
Author(s):  
Nicholas P. Luciano ◽  
Brian J. Ade ◽  
Kang Seog Kim ◽  
Andrew J. Conant
Keyword(s):  
Cross Section ◽  
Neutron Transport ◽  
Three Dimensional ◽  
Light Water Reactors ◽  
Pressurized Water Reactor ◽  
Reactor Physics ◽  
Pressurized Water ◽  
Water Reactors ◽  
Improved Performance ◽  
Burnup Calculation

MPACT is a state-of-the-art core simulator designed to perform high-fidelity analysis using whole-core, three-dimensional, pin-resolved neutron transport calculations on modern parallel computing hardware. MPACT was originally developed to model light water reactors, and its capabilities are being extended to simulate gas-cooled, graphite-moderated cores such as Magnox reactors. To verify MPACT’s performance in this new application, the code is being formally benchmarked using representative problems. Progression problems are a series of example models that increase in complexity designed to test a code’s performance. The progression problems include both beginning-of-cycle and depletion calculations. Reference solutions for each progression problem have been generated using Serpent 2, a continuous-energy Monte Carlo reactor physics burnup calculation code. Using the neutron multiplication eigenvalue ke_ as a metric, MPACT’s performance is assessed on each of the progression problems. Initial results showed that MPACT’s multigroup cross section libraries, originally developed for pressurized water reactor problems, were not sufficient to accurately solve Magnox problems. MPACT’s improved performance on the progression problems is demonstrated using this new optimized cross section library.

scholarly journals A Reconfigurable System Approach to the Direct Kinematics of a 5D.o.fRobotic Manipulator

2010 ◽  
Vol 2010 ◽  
pp. 1-10 ◽  
Author(s):  
Diego F. Sánchez ◽  
Daniel M. Muñoz ◽  
Carlos H. Llanos ◽  
José M. Motta
Keyword(s):  
High Performance ◽  
Degrees Of Freedom ◽  
Dynamic Range ◽  
Hardware Acceleration ◽  
Hardware Architecture ◽  
Floating Point ◽  
Field Programmable ◽  
Programmable Gate Arrays ◽  
Transcendental Functions ◽  
Direct Kinematics

Hardware acceleration in high performance computer systems has a particular interest for many engineering and scientific applications in which a large number of arithmetic operations and transcendental functions must be computed. In this paper a hardware architecture for computing direct kinematics of robot manipulators with 5 degrees of freedom (5D.o.f) using floating-point arithmetic is presented for 32, 43, and 64 bit-width representations and it is implemented in Field Programmable Gate Arrays (FPGAs). The proposed architecture has been developed using several floating-point libraries for arithmetic and transcendental functions operators, allowing the designer to select (pre-synthesis) a suitable bit-width representation according to the accuracy and dynamic range, as well as the area, elapsed time and power consumption requirements of the application. Synthesis results demonstrate the effectiveness and high performance of the implemented cores on commercial FPGAs. Simulation results have been addressed in order to compute the Mean Square Error (MSE), using the Matlab as statistical estimator, validating the correct behavior of the implemented cores. Additionally, the processing time of the hardware architecture was compared with the same formulation implemented in software, using the PowerPC (FPGA embedded processor), demonstrating that the hardware architecture speeds-up by factor of 1298 the software implementation.

scholarly journals Hardware Acceleration of High-Performance Computational Flow Dynamics Using High-Bandwidth Memory-Enabled Field-Programmable Gate Arrays

2022 ◽  
Vol 15 (2) ◽  
pp. 1-35
Author(s):  
Tom Hogervorst ◽  
Răzvan Nane ◽  
Giacomo Marchiori ◽  
Tong Dong Qiu ◽  
Markus Blatt ◽  
...  
Keyword(s):  
High Performance ◽  
Scientific Computing ◽  
Hardware Acceleration ◽  
Field Programmable Gate Arrays ◽  
Gate Arrays ◽  
Computational Flow Dynamics ◽  
Field Programmable ◽  
Programmable Gate Arrays ◽  
High Bandwidth ◽  
Reservoir Simulator

Scientific computing is at the core of many High-Performance Computing applications, including computational flow dynamics. Because of the utmost importance to simulate increasingly larger computational models, hardware acceleration is receiving increased attention due to its potential to maximize the performance of scientific computing. Field-Programmable Gate Arrays could accelerate scientific computing because of the possibility to fully customize the memory hierarchy important in irregular applications such as iterative linear solvers. In this article, we study the potential of using Field-Programmable Gate Arrays in High-Performance Computing because of the rapid advances in reconfigurable hardware, such as the increase in on-chip memory size, increasing number of logic cells, and the integration of High-Bandwidth Memories on board. To perform this study, we propose a novel Sparse Matrix-Vector multiplication unit and an ILU0 preconditioner tightly integrated with a BiCGStab solver kernel. We integrate the developed preconditioned iterative solver in Flow from the Open Porous Media project, a state-of-the-art open source reservoir simulator. Finally, we perform a thorough evaluation of the FPGA solver kernel in both stand-alone mode and integrated in the reservoir simulator, using the NORNE field, a real-world case reservoir model using a grid with more than 10 5 cells and using three unknowns per cell.

scholarly journals SERPENT/SUBCHANFLOW COUPLED CALCULATIONS FOR A VVER CORE AT HOT FULL POWER

2021 ◽  
Vol 247 ◽  
pp. 04006
Author(s):  
Diego Ferraro ◽  
Manuel García ◽  
Uwe Imke ◽  
Ville Valtavirta ◽  
Riku Tuominen ◽  
...  
Keyword(s):  
High Performance ◽  
Neutron Transport ◽  
Coupling Scheme ◽  
High Fidelity ◽  
Reactor Physics ◽  
Coupled Calculation ◽  
Final Goal ◽  
Wide Range ◽  
Computational Resources ◽  
Full Core

An increasing interest on the development of highly accurate methodologies in reactor physics is nowadays observed, mainly stimulated by the availability of vast computational resources. As a result, an on-going development of a wide range of coupled calculation tools is observed within diverse projects worldwide. Under this framework, the McSAFE European Union project is a coordinated effort aimed to develop multiphysics tools based on Monte Carlo neutron transport and subchannel thermal-hydraulics codes. These tools are aimed to be suitable for high-fidelity calculations both for PWR and VVER reactors, with the final goal of performing pin-by-pin coupled calculations at full core scope including burnup. Several intermediate steps are to be analyzed in-depth before jumping into this final goal in order to provide insights and to identify resources requirements. As part of this process, this work presents the results for a pin-by-pin coupling calculation using the Serpent 2 code (developed by VTT, Finland) and the subchannel code SUBCHANFLOW (SCF, developed by KIT, Germany) for a full-core VVER model. For such purpose, a recently refurbished master-slave coupling scheme is used within a High Performance Computing architecture. A full-core benchmark for a VVER-1000 that provides experimental data is considered, where the first burnup step (i.e. fresh core at hot-full rated power state) is calculated. For such purpose a detailed (i.e. pin-by-pin) coupled Serpent-SCF model is developed, including a simplified equilibrium xenon distribution (i.e. by fuel assembly). Comparisons with main global reported results are presented and briefly discussed, together with a raw estimation of resources requirements and a brief demonstration of the inherent capabilities of the proposed approach. The results presented here provide valuable insights and pave the way to tackle the final goals of the on-going high-fidelity project.

scholarly journals Martian dust devils detector over FPGA

2012 ◽  
Vol 1 (1) ◽  
pp. 23-31 ◽  
Author(s):  
E. de Lucas ◽  
M. J. Miguel ◽  
D. Mozos ◽  
L. Vázquez
Keyword(s):  
High Performance ◽  
Hardware Implementation ◽  
Small Volume ◽  
Space Missions ◽  
Reconfigurable Hardware ◽  
Dust Devil ◽  
Dust Devils ◽  
Gate Arrays ◽  
Field Programmable ◽  
Programmable Gate Arrays

Abstract. Digital applications that must be on-board space missions must comply with a very restrictive set of requirements. These include energy efficiency, small volume and weight, robustness and high performance. Moreover, these circuits cannot be repaired in case of error, so they must be reliable or provide some way to recover from errors. These features make reconfigurable hardware (FPGAs, Field Programmable Gate Arrays) a very suitable technology to be used in space missions. This paper presents a Martian dust devil detector implemented on an FPGA. The results show that a hardware implementation of the algorithm presents very good numbers in terms of performance compared with the software version. Moreover, as the amount of time needed to perform all the computations on the reconfigurable hardware is small, this hardware can be used most of the time to realize other applications.

scholarly journals TOWARDS CAD-BASED GEOMETRY MODELLING WITH THE RANDOM RAY METHOD

2021 ◽  
Vol 247 ◽  
pp. 03023
Author(s):  
Patrick C. Shriwise ◽  
John R. Tramm ◽  
Andrew Davis ◽  
Paul K. Romano
Keyword(s):  
Computer Aided Design ◽  
High Performance ◽  
Neutron Transport ◽  
Traditional Approach ◽  
Three Dimensional ◽  
Test Problems ◽  
Complex Geometries ◽  
Ray Method ◽  
Axial Section ◽  
Pressurized Water

The Advanced Random Ray Code (ARRC) is a high performance computing application capable of high-fidelity simulations of full core nuclear reactor models. ARRC leverages a recently developed stochastic method for neutron transport, known as The Random Ray Method (TRRM), which offers a variety of computational and numerical advantages as compared to existing methods. In particular, TRRM has been shown to be capable of efficient simulation of explicit three dimensional geometry representations without assumptions about axial homogeneity. To date, ARRC has utilized Constructive Solid Geometry (CSG) combined with a nested lattice geometry which works well for typical pressurized water reactors, but is not performant for the general case featuring arbitrary geometries. To facilitate simulation of arbitrarily complex geometries in ARRC efficiently, we propose performing transport directly on Computer-Aided Design (CAD) models of the geometry. In this study, we utilize the Direct-Accelerated Geometry Monte Carlo (DAGMC) toolkit which tracks particles on tessellated CAD geometries using a bounding volume hierarchy to accelerate the process, as a replacement for ARRC’s current lattice-based accelerations. Additionally, we present a method for automatically subdividing the large CAD regions in the DAGMC model into smaller mesh cells required by random ray to achieve high accuracy. We test the new DAGMC geometry implementation in ARRC on several test problems, including a 3D pincells, 3D assemblies, and an axial section of the Advanced Test Reactor. We show that DAGMC allows for simulation of complex geometries in ARRC that would otherwise not be possible using the traditional approach while maintaining solution accuracy.

Sign in / Sign up

Export Citation Format

Share Document